Tom Clegg [Tue, 15 Aug 2023 17:40:18 +0000 (13:40 -0400)]
17244: Add upgrade note re number formatting change.
Arvados-DCO-1.1-Signed-off-by: Tom Clegg <tom@curii.com>
Tom Clegg [Mon, 14 Aug 2023 19:40:36 +0000 (15:40 -0400)]
17244: New debian10 testdata, add cpu.max and cpuinfo for others.
Arvados-DCO-1.1-Signed-off-by: Tom Clegg <tom@curii.com>
Tom Clegg [Mon, 14 Aug 2023 19:09:27 +0000 (15:09 -0400)]
17244: Use cpu.max to get available fraction of CPUs.
Arvados-DCO-1.1-Signed-off-by: Tom Clegg <tom@curii.com>
Tom Clegg [Mon, 14 Aug 2023 20:29:48 +0000 (16:29 -0400)]
Merge branch '20649-ssh-help'
fixes #20649
Arvados-DCO-1.1-Signed-off-by: Tom Clegg <tom@curii.com>
Tom Clegg [Mon, 14 Aug 2023 20:29:09 +0000 (16:29 -0400)]
Merge branch '20457-careful-near-quota'
fixes #20457
Arvados-DCO-1.1-Signed-off-by: Tom Clegg <tom@curii.com>
Tom Clegg [Mon, 14 Aug 2023 15:14:24 +0000 (11:14 -0400)]
20649: Mention `arvados-server cloudtest` in boot timeout log msg.
Arvados-DCO-1.1-Signed-off-by: Tom Clegg <tom@curii.com>
Tom Clegg [Mon, 14 Aug 2023 15:11:51 +0000 (11:11 -0400)]
20649: Fix panic on race, worker shutdown vs. container startup.
Arvados-DCO-1.1-Signed-off-by: Tom Clegg <tom@curii.com>
Tom Clegg [Mon, 14 Aug 2023 15:10:44 +0000 (11:10 -0400)]
20649: Don't wait timeoutSignal before sending first TERM signal.
Arvados-DCO-1.1-Signed-off-by: Tom Clegg <tom@curii.com>
Lucas Di Pentima [Mon, 14 Aug 2023 14:07:19 +0000 (11:07 -0300)]
Fixes comment formatting. No issue #
Arvados-DCO-1.1-Signed-off-by: Lucas Di Pentima <lucas.dipentima@curii.com>
Tom Clegg [Wed, 9 Aug 2023 21:07:35 +0000 (17:07 -0400)]
20457: Fix flaky test.
Arvados-DCO-1.1-Signed-off-by: Tom Clegg <tom@curii.com>
Tom Clegg [Fri, 11 Aug 2023 21:47:32 +0000 (17:47 -0400)]
20649: Log actual error from last failed boot probe.
Not just the stderr content, which is useless if the error was
something like "connection failed".
Arvados-DCO-1.1-Signed-off-by: Tom Clegg <tom@curii.com>
Tom Clegg [Fri, 11 Aug 2023 21:47:06 +0000 (17:47 -0400)]
20649: Log boot probe errors other than usual wait-for-boot errors.
Arvados-DCO-1.1-Signed-off-by: Tom Clegg <tom@curii.com>
Tom Clegg [Fri, 11 Aug 2023 17:31:29 +0000 (13:31 -0400)]
20457: Update test to match new unlock-at-quota behavior.
Arvados-DCO-1.1-Signed-off-by: Tom Clegg <tom@curii.com>
Tom Clegg [Fri, 11 Aug 2023 14:17:30 +0000 (10:17 -0400)]
20457: Don't lock-cycle next-in-line ctr while waiting for quota.
Arvados-DCO-1.1-Signed-off-by: Tom Clegg <tom@curii.com>
Tom Clegg [Fri, 11 Aug 2023 14:12:25 +0000 (10:12 -0400)]
20457: Fix total instances count.
Arvados-DCO-1.1-Signed-off-by: Tom Clegg <tom@curii.com>
Peter Amstutz [Thu, 10 Aug 2023 18:24:53 +0000 (14:24 -0400)]
Add packages required to build newer singularity, refs #20735
Arvados-DCO-1.1-Signed-off-by: Peter Amstutz <peter.amstutz@curii.com>
Peter Amstutz [Thu, 10 Aug 2023 17:11:30 +0000 (13:11 -0400)]
Merge branch '20688-wb1-to-wb2-redirects' refs #20688
Arvados-DCO-1.1-Signed-off-by: Peter Amstutz <peter.amstutz@curii.com>
Tom Clegg [Thu, 10 Aug 2023 14:45:46 +0000 (10:45 -0400)]
20649: cloudtest obeys DeployPublicKey config, like a-d-c.
Arvados-DCO-1.1-Signed-off-by: Tom Clegg <tom@curii.com>
Tom Clegg [Thu, 10 Aug 2023 14:34:11 +0000 (10:34 -0400)]
Merge branch '20755-ec2-multiple-subnets'
closes #20755
Arvados-DCO-1.1-Signed-off-by: Tom Clegg <tom@curii.com>
Tom Clegg [Wed, 9 Aug 2023 21:07:35 +0000 (17:07 -0400)]
20457: Fix flaky test.
Arvados-DCO-1.1-Signed-off-by: Tom Clegg <tom@curii.com>
Tom Clegg [Wed, 9 Aug 2023 21:06:40 +0000 (17:06 -0400)]
20457: Exercise quota handling in dispatcher chaos test.
Arvados-DCO-1.1-Signed-off-by: Tom Clegg <tom@curii.com>
Tom Clegg [Mon, 7 Aug 2023 19:26:20 +0000 (15:26 -0400)]
20457: When passing prior quota level, raise maxConcurrency slowly.
Arvados-DCO-1.1-Signed-off-by: Tom Clegg <tom@curii.com>
Peter Amstutz [Wed, 9 Aug 2023 13:53:48 +0000 (09:53 -0400)]
Merge branch '20680-rolling-deploy' refs #20680
Arvados-DCO-1.1-Signed-off-by: Peter Amstutz <peter.amstutz@curii.com>
Peter Amstutz [Wed, 9 Aug 2023 13:48:51 +0000 (09:48 -0400)]
20680: Increase MaxConcurrentRequests to 8
Necessary to avoid a API -> keep-web -> API loop causing a deadlock.
Arvados-DCO-1.1-Signed-off-by: Peter Amstutz <peter.amstutz@curii.com>
Tom Clegg [Wed, 9 Aug 2023 13:33:34 +0000 (09:33 -0400)]
20755: Merge branch 'main' into 20755-ec2-multiple-subnets
Arvados-DCO-1.1-Signed-off-by: Tom Clegg <tom@curii.com>
Tom Clegg [Wed, 9 Aug 2023 13:32:37 +0000 (09:32 -0400)]
20755: Fix test case by accepting empty DriverParameters.
Arvados-DCO-1.1-Signed-off-by: Tom Clegg <tom@curii.com>
Tom Clegg [Tue, 8 Aug 2023 21:31:24 +0000 (17:31 -0400)]
20755: Run cloudtest suite once per configured subnet, not just 1st.
Arvados-DCO-1.1-Signed-off-by: Tom Clegg <tom@curii.com>
Tom Clegg [Tue, 8 Aug 2023 18:52:42 +0000 (14:52 -0400)]
Merge branch '20735-update-deps'
refs #20735
Arvados-DCO-1.1-Signed-off-by: Tom Clegg <tom@curii.com>
Tom Clegg [Tue, 8 Aug 2023 18:51:21 +0000 (14:51 -0400)]
20735: Clarify singularity version/compatibility expectations.
Arvados-DCO-1.1-Signed-off-by: Tom Clegg <tom@curii.com>
Peter Amstutz [Tue, 8 Aug 2023 17:42:25 +0000 (13:42 -0400)]
20680: Only do a rolling update when there is multiple controllers
Arvados-DCO-1.1-Signed-off-by: Peter Amstutz <peter.amstutz@curii.com>
Tom Clegg [Tue, 8 Aug 2023 14:18:29 +0000 (10:18 -0400)]
Fix error check.
refs #17244
Arvados-DCO-1.1-Signed-off-by: Tom Clegg <tom@curii.com>
Peter Amstutz [Tue, 8 Aug 2023 13:36:15 +0000 (09:36 -0400)]
20680: installer does rolling update of controller
When there is a load balancer, disable each controller node in turn
before updating it.
Also, don't reduce controller_nr when a controller is down, having
worker_connections go up and down during deploy seems like in could
cause problems.
Arvados-DCO-1.1-Signed-off-by: Peter Amstutz <peter.amstutz@curii.com>
Brett Smith [Mon, 7 Aug 2023 21:18:56 +0000 (17:18 -0400)]
Merge branch '20754-docker-py-upgrade'
Closes #20754.
Arvados-DCO-1.1-Signed-off-by: Brett Smith <brett.smith@curii.com>
Tom Clegg [Mon, 7 Aug 2023 20:58:45 +0000 (16:58 -0400)]
20735: Merge branch 'main' into 20735-update-deps
Arvados-DCO-1.1-Signed-off-by: Tom Clegg <tom@curii.com>
Tom Clegg [Mon, 7 Aug 2023 20:57:23 +0000 (16:57 -0400)]
20735: Update singularity version in docs and build script.
Arvados-DCO-1.1-Signed-off-by: Tom Clegg <tom@curii.com>
Tom Clegg [Mon, 7 Aug 2023 20:49:17 +0000 (16:49 -0400)]
Merge branch '20705-crunchstat-warn-missing-data'
fixes #20705
Arvados-DCO-1.1-Signed-off-by: Tom Clegg <tom@curii.com>
Peter Amstutz [Mon, 7 Aug 2023 20:29:15 +0000 (16:29 -0400)]
Merge branch '20680-default-config-updates' refs #20680
Arvados-DCO-1.1-Signed-off-by: Peter Amstutz <peter.amstutz@curii.com>
Peter Amstutz [Mon, 7 Aug 2023 20:28:13 +0000 (16:28 -0400)]
20680: Fix nginx_balancer_configuration & add note about defaults
Arvados-DCO-1.1-Signed-off-by: Peter Amstutz <peter.amstutz@curii.com>
Tom Clegg [Mon, 7 Aug 2023 14:31:56 +0000 (10:31 -0400)]
20705: 10% failure on stub Create to ensure running error path.
Fixes occasional test failures due to absence of rate-limiting errors.
Arvados-DCO-1.1-Signed-off-by: Tom Clegg <tom@curii.com>
Tom Clegg [Mon, 7 Aug 2023 14:04:38 +0000 (10:04 -0400)]
20735: Sync bundler version.
See https://dev.arvados.org/issues/20735#note-6
Arvados-DCO-1.1-Signed-off-by: Tom Clegg <tom@curii.com>
Lucas Di Pentima [Fri, 4 Aug 2023 20:55:13 +0000 (17:55 -0300)]
Fixes instance profile assignment. Refs #20610
Arvados-DCO-1.1-Signed-off-by: Lucas Di Pentima <lucas.dipentima@curii.com>
Lucas Di Pentima [Fri, 4 Aug 2023 19:08:32 +0000 (16:08 -0300)]
Merge branch '20610-installer-load-balancer'. Refs #20610
Arvados-DCO-1.1-Signed-off-by: Lucas Di Pentima <lucas.dipentima@curii.com>
Brett Smith [Fri, 4 Aug 2023 18:28:18 +0000 (14:28 -0400)]
20754: Upgrade docker module for docker-cleaner
The immediate motivation is dealing with the bug discussed in the
comments. Upgrading generally seems like a good idea since we've been
pinned at a specific version for so long.
Changing the Docker API version is required by the library: 1.21 is the
oldest version currently supported. Our Go stack is standardized on
1.35, so use that here too.
Arvados-DCO-1.1-Signed-off-by: Brett Smith <brett.smith@curii.com>
Lucas Di Pentima [Fri, 4 Aug 2023 18:08:04 +0000 (15:08 -0300)]
20610: Documentation changes, including simplification & fixes.
Arvados-DCO-1.1-Signed-off-by: Lucas Di Pentima <lucas.dipentima@curii.com>
Brett Smith [Fri, 4 Aug 2023 18:05:26 +0000 (14:05 -0400)]
20754: Remove test requirements from docker-cleaner
Neither are required with our current Python stack.
Arvados-DCO-1.1-Signed-off-by: Brett Smith <brett.smith@curii.com>
Lucas Di Pentima [Fri, 4 Aug 2023 17:48:29 +0000 (14:48 -0300)]
20610: Changes the default installer config in order to simplify documentation.
Arvados-DCO-1.1-Signed-off-by: Lucas Di Pentima <lucas.dipentima@curii.com>
Peter Amstutz [Thu, 3 Aug 2023 14:31:05 +0000 (10:31 -0400)]
20688: Update single host/multiple hostname template
Arvados-DCO-1.1-Signed-off-by: Peter Amstutz <peter.amstutz@curii.com>
Peter Amstutz [Thu, 3 Aug 2023 13:44:41 +0000 (09:44 -0400)]
20688: "workbench" host is now workbench2
"workbench2" host redirects to workbench
Right now, both workbench and workbench2 roles still need to be
assigned, because the 'workbench' role is now configuring nginx but
the 'workbench2' state is the one that actually installs the package.
Will need to make some upstream changes to the formula.
Arvados-DCO-1.1-Signed-off-by: Peter Amstutz <peter.amstutz@curii.com>
Peter Amstutz [Wed, 2 Aug 2023 21:18:32 +0000 (17:18 -0400)]
20688: Reorder redirects by priority
Arvados-DCO-1.1-Signed-off-by: Peter Amstutz <peter.amstutz@curii.com>
Peter Amstutz [Wed, 2 Aug 2023 18:39:20 +0000 (14:39 -0400)]
20688: check for ssh ahead of time, sync only before deploy
Arvados-DCO-1.1-Signed-off-by: Peter Amstutz <peter.amstutz@curii.com>
Peter Amstutz [Fri, 21 Jul 2023 17:44:36 +0000 (13:44 -0400)]
20688: Add nginx redirects from wb1 paths to wb2 paths to salt
Arvados-DCO-1.1-Signed-off-by: Peter Amstutz <peter.amstutz@curii.com>
Peter Amstutz [Fri, 21 Jul 2023 17:15:11 +0000 (13:15 -0400)]
20688: Fix quoting issue
Arvados-DCO-1.1-Signed-off-by: Peter Amstutz <peter.amstutz@curii.com>
Peter Amstutz [Tue, 18 Jul 2023 18:57:39 +0000 (14:57 -0400)]
20688: Add wb1 to wb2 redirects to arvbox
Arvados-DCO-1.1-Signed-off-by: Peter Amstutz <peter.amstutz@curii.com>
Peter Amstutz [Thu, 3 Aug 2023 18:26:51 +0000 (14:26 -0400)]
20680: Rework worker settings for new controller behavior
We now have separate values for concurrent requests and queued
requests. Ensure the arvados, passenger and nginx configurations align
with the correct values.
Renamed CONTROLLER_NGINX_WORKERS to CONTROLLER_MAX_WORKERS
Renamed CONTROLLER_MAX_CONCURRENT_REQUESTS to
CONTROLLER_MAX_QUEUED_REQUESTS
Adjusted config.yml defaults to reflect that MaxConcurrentRequests
means something different.
Arvados-DCO-1.1-Signed-off-by: Peter Amstutz <peter.amstutz@curii.com>
Tom Clegg [Fri, 4 Aug 2023 14:46:00 +0000 (10:46 -0400)]
20755: Report metrics for ec2 instances per subnet.
Arvados-DCO-1.1-Signed-off-by: Tom Clegg <tom@curii.com>
Lucas Di Pentima [Thu, 3 Aug 2023 20:53:26 +0000 (17:53 -0300)]
20610: Documentation fixes & additions.
Also, removed the 'api' role from the multi host local.params example.
Arvados-DCO-1.1-Signed-off-by: Lucas Di Pentima <lucas.dipentima@curii.com>
Lucas Di Pentima [Thu, 3 Aug 2023 19:58:18 +0000 (16:58 -0300)]
20610: Fixes balancer nginx's request queue config depending on # of backends
Also, fixes templating issue on arvados' config.yml file related to this
setting.
Arvados-DCO-1.1-Signed-off-by: Lucas Di Pentima <lucas.dipentima@curii.com>
Tom Clegg [Thu, 3 Aug 2023 19:50:50 +0000 (15:50 -0400)]
20755: Allow cloud drivers to register their own metrics.
Arvados-DCO-1.1-Signed-off-by: Tom Clegg <tom@curii.com>
Tom Clegg [Thu, 3 Aug 2023 19:03:51 +0000 (15:03 -0400)]
20755: Support multiple/alternate subnets on EC2.
Arvados-DCO-1.1-Signed-off-by: Tom Clegg <tom@curii.com>
Lucas Di Pentima [Thu, 3 Aug 2023 18:51:13 +0000 (15:51 -0300)]
20610: Unifies the 'api' role with the 'controller' role.
Arvados-DCO-1.1-Signed-off-by: Lucas Di Pentima <lucas.dipentima@curii.com>
Lucas Di Pentima [Thu, 3 Aug 2023 15:06:51 +0000 (12:06 -0300)]
20610: Improves provision.sh code readability.
Arvados-DCO-1.1-Signed-off-by: Lucas Di Pentima <lucas.dipentima@curii.com>
Lucas Di Pentima [Thu, 3 Aug 2023 14:05:24 +0000 (11:05 -0300)]
20610: Deduplicates variable substitution code.
Arvados-DCO-1.1-Signed-off-by: Lucas Di Pentima <lucas.dipentima@curii.com>
Lucas Di Pentima [Wed, 2 Aug 2023 21:21:22 +0000 (18:21 -0300)]
20610: Renames the ROLES map variable because the name was being used already.
Also, adds "set -eu" to provision.sh and fixes related unbound var issues.
Arvados-DCO-1.1-Signed-off-by: Lucas Di Pentima <lucas.dipentima@curii.com>
Tom Clegg [Wed, 2 Aug 2023 20:11:11 +0000 (16:11 -0400)]
20735: Update Go to 1.20.6.
Arvados-DCO-1.1-Signed-off-by: Tom Clegg <tom@curii.com>
Tom Clegg [Wed, 12 Jul 2023 15:35:48 +0000 (11:35 -0400)]
20735: Update ruby and singularity.
Arvados-DCO-1.1-Signed-off-by: Tom Clegg <tom@curii.com>
Tom Clegg [Wed, 12 Jul 2023 15:34:45 +0000 (11:34 -0400)]
20735: Update to go 1.20.
Arvados-DCO-1.1-Signed-off-by: Tom Clegg <tom@curii.com>
Lucas Di Pentima [Wed, 2 Aug 2023 13:38:07 +0000 (10:38 -0300)]
20610: Moves code from local.params to its own common.sh file.
provision & installer scripts now loads this new file that in turn loads
the params files.
Arvados-DCO-1.1-Signed-off-by: Lucas Di Pentima <lucas.dipentima@curii.com>
Tom Clegg [Wed, 2 Aug 2023 13:24:32 +0000 (09:24 -0400)]
Merge branch '17244-cgroup2'
refs #17244
Arvados-DCO-1.1-Signed-off-by: Tom Clegg <tom@curii.com>
Tom Clegg [Wed, 2 Aug 2023 13:23:40 +0000 (09:23 -0400)]
17244: Fix doc link.
Arvados-DCO-1.1-Signed-off-by: Tom Clegg <tom@curii.com>
Brett Smith [Wed, 2 Aug 2023 01:08:36 +0000 (21:08 -0400)]
Merge branch '20797-rhel8-packages'
Closes #20797.
Arvados-DCO-1.1-Signed-off-by: Brett Smith <brett.smith@curii.com>
Brett Smith [Tue, 1 Aug 2023 18:07:32 +0000 (14:07 -0400)]
20797: Merge Red Hat package install instructions
For now, since the instructions are *exactly* the same across distros
and versions, this makes sense as a streamlining measure. If we add more
variation to the process in the future, it probably makes sense to split
these out again.
Arvados-DCO-1.1-Signed-off-by: Brett Smith <brett.smith@curii.com>
Brett Smith [Tue, 1 Aug 2023 16:06:59 +0000 (12:06 -0400)]
20797: Run microdnf with --assumeyes
`docker run --rm -ti rockylinux:8-minimal microdnf --help` says,
> The "--assumeyes" option is turned on by default.
However, that behavior apparently changes in a future version:
<https://github.com/rpm-software-management/microdnf/pull/97>,
microdnf commit
ca778a0b34c5e15b5b0de1e0af0f8892280fc0c2.
Passing `--assumeyes` provides more consistency and will help keep the
code working in later releases.
Arvados-DCO-1.1-Signed-off-by: Brett Smith <brett.smith@curii.com>
Brett Smith [Tue, 1 Aug 2023 16:02:30 +0000 (12:02 -0400)]
20797: Prefer dnf in PostgreSQL install instructions
I believe this is the current preferred tool, and provides consistency
with other RH8 documentation.
Arvados-DCO-1.1-Signed-off-by: Brett Smith <brett.smith@curii.com>
Tom Clegg [Mon, 31 Jul 2023 19:52:50 +0000 (15:52 -0400)]
17244: Update docs for -cgroup-parent-subsystem flag.
Arvados-DCO-1.1-Signed-off-by: Tom Clegg <tom@curii.com>
Tom Clegg [Mon, 31 Jul 2023 19:33:18 +0000 (15:33 -0400)]
17244: Clarify log message.
Arvados-DCO-1.1-Signed-off-by: Tom Clegg <tom@curii.com>
Tom Clegg [Mon, 31 Jul 2023 17:54:24 +0000 (13:54 -0400)]
20705: Warn if some stats are missing.
Arvados-DCO-1.1-Signed-off-by: Tom Clegg <tom@curii.com>
Brett Smith [Mon, 31 Jul 2023 15:37:38 +0000 (11:37 -0400)]
20797: Update cross compilation detection logic
The immediate bug that needs to be fixed is that we currently don't
support cross compilation on rocky8.
DRY up the code to try to make it easier to follow what's going on
here.
Arvados-DCO-1.1-Signed-off-by: Brett Smith <brett.smith@curii.com>
Brett Smith [Fri, 28 Jul 2023 19:53:45 +0000 (15:53 -0400)]
20797: Note Alma/CentOS/Red Hat/Rocky 8 support in install docs
Arvados-DCO-1.1-Signed-off-by: Brett Smith <brett.smith@curii.com>
Brett Smith [Fri, 28 Jul 2023 19:12:44 +0000 (15:12 -0400)]
20797: Copyedit "CentOS" for consistent capitalization
I left the upgrade notes alone since they're more anchored in time, but
this should get the rest of the documentation.
Arvados-DCO-1.1-Signed-off-by: Brett Smith <brett.smith@curii.com>
Brett Smith [Fri, 28 Jul 2023 18:51:18 +0000 (14:51 -0400)]
20797: Add basic documentation for how to add a new target distro
Arvados-DCO-1.1-Signed-off-by: Brett Smith <brett.smith@curii.com>
Brett Smith [Fri, 28 Jul 2023 18:41:15 +0000 (14:41 -0400)]
20797: Add basic intro documentation to build README
Arvados-DCO-1.1-Signed-off-by: Brett Smith <brett.smith@curii.com>
Brett Smith [Fri, 28 Jul 2023 19:32:32 +0000 (15:32 -0400)]
20797: Update bundler version in rocky8 Docker images
This updates the version to match our `Gemfile.lock` files and avoids a
warning.
Arvados-DCO-1.1-Signed-off-by: Brett Smith <brett.smith@curii.com>
Brett Smith [Fri, 28 Jul 2023 19:01:17 +0000 (15:01 -0400)]
20797: Update for new createrepo package+executable name
In Debian 12 the package is named createrepo-c and it installs
/usr/bin/createrepo_c. With this change, the build script prefers that
name, but still looks for and uses `createrepo`.
Arvados-DCO-1.1-Signed-off-by: Brett Smith <brett.smith@curii.com>
Brett Smith [Fri, 28 Jul 2023 18:09:02 +0000 (14:09 -0400)]
20797: Streamline setup of Python build variables
Instead of setting Debian defaults and overriding them,
1. set global defaults
2. set target-specific variables
3. set deb/rpm-specific defaults as needed
I hope this makes clearer what's actually a target-specific variable
vs. a regular difference between Debian and Red Hat, and so makes it
easier to add targets in the future.
Arvados-DCO-1.1-Signed-off-by: Brett Smith <brett.smith@curii.com>
Brett Smith [Fri, 28 Jul 2023 17:48:08 +0000 (13:48 -0400)]
20797: Add initial support for rocky8 package builds
Arvados-DCO-1.1-Signed-off-by: Brett Smith <brett.smith@curii.com>
Lucas Di Pentima [Fri, 28 Jul 2023 14:43:39 +0000 (11:43 -0300)]
20610: Updates docs to reflect changes on arvados.sls configuration.
Arvados-DCO-1.1-Signed-off-by: Lucas Di Pentima <lucas.dipentima@curii.com>
Lucas Di Pentima [Fri, 28 Jul 2023 14:32:07 +0000 (11:32 -0300)]
20610: Adds documentation for load-balancing & rolling upgrades.
Arvados-DCO-1.1-Signed-off-by: Lucas Di Pentima <lucas.dipentima@curii.com>
Lucas Di Pentima [Fri, 28 Jul 2023 13:48:40 +0000 (10:48 -0300)]
20610: Removes the need to manually set up the ENABLE_BALANCER variable.
Having the role->nodes map allows us to simplify manual configuration.
Arvados-DCO-1.1-Signed-off-by: Lucas Di Pentima <lucas.dipentima@curii.com>
Lucas Di Pentima [Thu, 27 Jul 2023 21:17:09 +0000 (18:17 -0300)]
20610: Fixes deployment order to avoid failures.
Arvados-DCO-1.1-Signed-off-by: Lucas Di Pentima <lucas.dipentima@curii.com>
Lucas Di Pentima [Thu, 27 Jul 2023 19:32:36 +0000 (16:32 -0300)]
20610: Allows disabling backend controllers for rolling updates.
Arvados-DCO-1.1-Signed-off-by: Lucas Di Pentima <lucas.dipentima@curii.com>
Lucas Di Pentima [Thu, 27 Jul 2023 18:39:56 +0000 (15:39 -0300)]
20610: Restricts backends' HTTP access by nginx rules.
I think this is better than implementing those controls through security
groups via Terraform, because the node's role information is already available
on the salt code.
Arvados-DCO-1.1-Signed-off-by: Lucas Di Pentima <lucas.dipentima@curii.com>
Lucas Di Pentima [Wed, 26 Jul 2023 22:43:57 +0000 (19:43 -0300)]
20610: Fixes prometheus configuration to track the controller nodes.
Also, fixes a-d-c config, as it was also assumed to be on the controller node.
Arvados-DCO-1.1-Signed-off-by: Lucas Di Pentima <lucas.dipentima@curii.com>
Lucas Di Pentima [Wed, 26 Jul 2023 21:06:41 +0000 (18:06 -0300)]
20610: Removes duplication on balancer & backend declarations.
Instead of having to explicitly declare those, just use the ROLES map
to get the balancer and controller nodes.
Also, explicitly gives PG access to Keep-Balance instead of assuming it's
installed on the same node as the controller.
Arvados-DCO-1.1-Signed-off-by: Lucas Di Pentima <lucas.dipentima@curii.com>
Lucas Di Pentima [Wed, 26 Jul 2023 19:57:09 +0000 (16:57 -0300)]
20610: Adds node list & role->nodes map variables.
Uses the node list variable to configure prometheus node exporter from the
configuration in local.params.
Arvados-DCO-1.1-Signed-off-by: Lucas Di Pentima <lucas.dipentima@curii.com>
Lucas Di Pentima [Wed, 26 Jul 2023 13:53:31 +0000 (10:53 -0300)]
20610: Fixes a template variable usage.
Arvados-DCO-1.1-Signed-off-by: Lucas Di Pentima <lucas.dipentima@curii.com>
Lucas Di Pentima [Wed, 26 Jul 2023 13:51:25 +0000 (10:51 -0300)]
20610: Removes the need to directly edit the arvados.sls pillar.
All params are now in local.params.* files.
Arvados-DCO-1.1-Signed-off-by: Lucas Di Pentima <lucas.dipentima@curii.com>
Brett Smith [Wed, 26 Jul 2023 13:40:39 +0000 (09:40 -0400)]
Merge branch '20543-pip-install-virtualenv'
Closes #20543.
Arvados-DCO-1.1-Signed-off-by: Brett Smith <brett.smith@curii.com>
Brett Smith [Fri, 7 Jul 2023 19:44:03 +0000 (15:44 -0400)]
20543: Recommend users install Python packages in a virtualenv
`pip install --user` doesn't work on distributions that have adopted
PEP 668, including Debian 12+ and Ubuntu 23.04+.
Arvados-DCO-1.1-Signed-off-by: Brett Smith <brett.smith@curii.com>
Brett Smith [Mon, 24 Jul 2023 14:53:13 +0000 (10:53 -0400)]
Merge branch '20343-regexp-prefixes'
Closes #20343.
Arvados-DCO-1.1-Signed-off-by: Brett Smith <brett.smith@curii.com>
Tom Clegg [Fri, 21 Jul 2023 23:05:33 +0000 (19:05 -0400)]
17244: Use crunch-run's cgroup, not init's, for "host" stats.
This gets much closer to "everything on the host" with minimal
effort.
Tracking the furthest ancestor of crunch-run's cgroup might be even
better, but will require additional support in crunchstat.
Arvados-DCO-1.1-Signed-off-by: Tom Clegg <tom@curii.com>
projects / arvados.git / log